def parseValues(self):
'''
Parses all data lines in file and yields them as a generator
'''
for i, dataLine in enumerate(self.mutationsFile):
if i > 1:
lineParts = dataLine.split(self.columnDelimiter)
hugoSymbol, entrezID, chrom, start, stop, strand, variantRef, refAllele, firstTumorSeqAllele, secondTumorSeqAllele, barcode = Utility.unpack_fromPositions(
lineParts, (0, 1, 4, 5, 6, 7, 9, 10, 11, 12, 15))
yield hugoSymbol, entrezID, chrom, start, stop, strand, variantRef, refAllele, firstTumorSeqAllele, secondTumorSeqAllele, barcode
def openFileAndParseHeaders(self):
'''
Opens the methylation array file in TCGA format and returns header values
'''
self.mutationsFile = open(self.fileName, 'r')
assert self.mutationsFile
print "Successfully accessed file: %s" % (self.mutationsFile)
for i, line in enumerate(self.mutationsFile):
if i == 0:
idHeader = line
elif i > 0:
break
idHeaderParts = idHeader.split(self.columnDelimiter)
hugoSymbol, entrezID, chrom, start, stop, strand, variantRef, refAllele, firstTumorSeqAllele, secondTumorSeqAllele = Utility.unpack_fromPositions(
idHeaderParts, (0, 1, 4, 5, 6, 7, 9, 10, 11, 12))
return hugoSymbol, entrezID, chrom, start, stop, strand, variantRef, refAllele, firstTumorSeqAllele, secondTumorSeqAllele